Agricultural Soil Science in a Changing Climate: Soil Health, Carbon Cycling, Nutrient Dynamics and Agroecosystem Resilience

Carbon farming has moved from an agronomic aspiration to a regulated and traded climate instrument, and the credibility of that transition rests on whether soil organic carbon change can be quantified accurately, attributed correctly and defended over time. Digital technologies, including proximal and laboratory spectroscopy, satellite earth observation, in-field sensor networks, process-based biogeochemical models, statistical learning and distributed registry infrastructure, are being assembled into digital monitoring, reporting and verification systems that promise to reduce cost while preserving rigour. This review critically evaluates the scientific basis of that promise across three connected domains: the biogeochemical limits of soil carbon accrual, the meaning and management of permanence, and the design and performance of digital quantification systems. Literature was identified through structured searching of open scholarly databases and indexes, supplemented by backward and forward citation searching and by targeted retrieval of institutional and policy analyses, with critical appraisal focused on measurement validity, uncertainty treatment, external validity and the alignment between evidence and claim. The synthesis indicates that the strongest evidence supports three positions: that measurable and durable gains in soil organic carbon are achievable but bounded by mineralogical capacity and by the diminishing returns of stock recovery; that reversibility is an intrinsic property of the resource rather than a project management failure, so accounting must treat impermanence explicitly; and that no single digital modality currently satisfies the joint requirements of accuracy, spatial coverage and affordability. Evidence is weakest, and disagreement most pronounced, in three areas: the transfer of spectroscopic and remote sensing calibrations to new soils and conditions, the independent temporal validation of the models used to interpolate between sampling events, and the empirical performance of digital systems in smallholder and dryland settings. Digital innovation improves the resolution and cost profile of quantification, but it does not by itself resolve questions of additionality, baseline construction or durability, which remain governance problems informed by, rather than solved by, measurement.